[Devel] [RFC][for -mm] memory controller enhancements for reclaiming take2 [4/8] possible race fix in res_counter

[KAMEZAWA Hiroyuki]

spinlock is necessary when someone changes res->counter value.
splited out from YAMAMOTO's background page reclaim for memory cgroup set.

Changelog v1 -> v2:
 - fixed type of "flags".


Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu at jp.fujitsu.com>
From: YAMAMOTO Takashi <yamamoto at valinux.co.jp>


 kernel/res_counter.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Index: linux-2.6.24-rc3-mm2/kernel/res_counter.c
===================================================================
--- linux-2.6.24-rc3-mm2.orig/kernel/res_counter.c
+++ linux-2.6.24-rc3-mm2/kernel/res_counter.c
@@ -98,6 +98,7 @@ ssize_t res_counter_write(struct res_cou
 {
 	int ret;
 	char *buf, *end;
+	unsigned long flags;
 	unsigned long long tmp, *val;
 
 	buf = kmalloc(nbytes + 1, GFP_KERNEL);
@@ -121,9 +122,10 @@ ssize_t res_counter_write(struct res_cou
 		if (*end != '\0')
 			goto out_free;
 	}
-
+	spin_lock_irqsave(&counter->lock, flags);
 	val = res_counter_member(counter, member);
 	*val = tmp;
+	spin_unlock_irqrestore(&counter->lock, flags);
 	ret = nbytes;
 out_free:
 	kfree(buf);

_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers